Specimen AICL — independent AI product studio

We build the AI products we wish existed.

AI Crafts Lab is a small studio that designs, builds, and ships AI products, tools, and automation — usually for ourselves first, then for whoever needs the same thing.

What we build

Five working modes, one studio.

AI products

Full products with a model at the core, built to be used daily, not demoed once.

AI tools

Focused, single-purpose tools that do one job well and stay out of the way.

Automation

Workflows that quietly remove a repeated task, and know when to hand back to a person.

Digital experiences

Interfaces and sites designed with the same care as the systems behind them.

Experiments

Small, fast prototypes built to answer one question — most are never shipped, and that's fine.

The lab

What's on the bench.

A running catalog of what we're building — some public, some still on the bench. Every entry gets a number and stays.

No. 001

Field notes

An internal research log where every experiment gets written up, win or fail. The first product built for the lab itself.

Internal toolIn progress
No. 002

Signal

A small automation that watches a workflow for the moment it needs a human, and only then.

AutomationExploring
No. 003

Untitled experiment

Early-stage. We build in the open once there's something worth showing.

ExperimentExploring

Philosophy

How we decide what to build.

Build useful things.
Not impressive things, not clever things — things a person keeps coming back to because it earns its place.
Experiment quickly.
Most ideas are wrong. The fastest way to find the right one is to build the cheap, ugly version first.
Combine AI, design, and engineering.
A model is not a product. The product is the judgment about where it helps and where it should stay quiet.
Turn ideas into working software.
A deck is a promise. Software is the thing that gets kept — so that's what we spend our time on.

Get in touch

Have something worth building? Tell us about it.

No pitch decks required. A clear problem and a rough idea of "done" is plenty to start with.